872 Appendix II. I "' ;* the screw taken at f- the I.H.P. The direct thrust may be found from speed, for V of ship = knots x 101*3 Thrust x knots x 101-3 = effective H.P. x 33000 I Qr p = (eff.H.P.)x 33000 ^LHjP, K x Tor3 K - f . , 217 I.H.P I.H.P. .-. Collar surface required =-------~ = 4*34 50 JtL jL It must be noted that these surfaces' are now horseshoe in form (see p. 691) and more collars are required than if circular. For pivot bearings the experimental apparatus in Fig. 833 was adopted, where B is the pivot or footstep fed by oil entering at pipe H. The shaft D being rotated through bevel gearing A A, the frictional moment was measured by weight G, which, acting on pulley F, prevented the bearing at B from rotating. At the same time, the load was obtained by oil pumped against surfaces D and E, its intensity being measured by pressure gauge c. P. 568. Balls arid Live Rollers.It is found that with ball or roller bearings the frictional loss is £ or $ of that of a plain journal, and in large bearings the rollers are kept apart by. rings, as in Fig. 5 84. P.575. Efficiencies of 'Machines.The example on pp. 571 to 575 shews the methods usually followed in mecha- nical laboratories to find frictional loss in machines. Two further cases may be given by way of illustration. Fig. 834 is a chart of experiments on rope-pulley blocks, a simple fixed pulley being called a ; i : i system, one movable and one fixed block a 2:1 system, and so on. Thus 3 upper and 3 lower pulleys make a 6: i system. Plotting P : W the inclined lines are obtained, and efficiency curves are further calculated for each load, A curve of maximum efficiency is also drawn belgw, on a base of velocity-ratio, which usefully indicates the rapid loss with increased theoretical advantage. The sheaves were 2§" diameter and the rope J". * Investigating, we have in a i: i system : P = W(r +m) where m is a proper fraction.